Currently building images on older kernels will fail because mkfs.btrfs
enables an incompatible feature 'extref' by default. We never really
made this requirement explicit and the SDK in general has continued to
maintain compatibility with older kernels. Make the requirement explicit
so users will get errors quicker and there is a clear line for what
kernel features can be used in the SDK.
Newer git ebuilds have decided that the "git-prompt" script isn't really
bash completion so stopped installing it via that mechanism. Instead it
installed it started installing it in /usr/share/docs which gets
compressed by default and the path is based on ebuild version. The path
changed again in 1.9.3 to /usr/share/git and didn't compress it so that
makes it actually possibly usable but 1.9.3 or later isn't stable yet.
We can re-enable it the next time git gets updated but not worth fussing
over the current brokenness right now.
Using parallel_emerge has been disabled by default for all commands
except build_image for quite a while now, build_image kept it just
because it was still a bit faster than normal emerge. Keeping
parallel_emerge complicates future changes to build_image so it needs to
drop it entirely. Since that means nothing uses it by default we might
as well just rip out support for it entirely.
The host system's PATH may not be match the one required by the SDK.
When going through the enter_chroot script it gets reset because bash is
invoked as a login shell but this doesn't happen when using the plain
old chroot command.
Fixes https://github.com/coreos/scripts/pull/290
The main case here is /etc/hosts does not exist on CoreOS. In the
process combine related and duplicate code. Setting the timezone now
happens in entire_chroot like hosts and resolv.conf. Don't bother with
setting a default UTC time zone, that is already the default.
To behave more like setup_board/build_packages update_chroot should
fully configure portage to make sure everything is accurate.
Now binhosts are defined in make.conf.host_setup so the static config in
coreos-overlays doesn't need to refer to version.txt. setup_board
already made this change in 7a43a07f.
Define path locations to reduce dependency between static configs in
coreos-overlays and the behavior of the scripts repo. Spreading
configuration across two repos makes everything harder to understand.
Eventually everything should either be defined in profiles in
coreos-overlays or minimal auto-generated config files here in scripts.
For the most part this doesn't influence anything. The one exception is
the custom configuration for using curl is dropped, just rely on the
portage defaults. It appears curl was only used to work around a wget
issue with Google's internal SSL certificates. We care not. :)
The commands useradd/usermod will silently skip adding users to
secondary groups that are not in /etc/group. The idea being that the
tools should not create groups that conflict with existing LDAP/NIS
groups but why trying to do so isn't a fatal error I don't know.
Overall the code is rather complicated and tries to modify instead of
add when possible to allow running the SDK as the 'core' user. To keep
things simple gut this code, make the 'core' user special, and add
secondary groups via the 'gpasswd' command so that errors are reported
instead of silently ignored.
One functional change: the default groups have changed to kvm and
portage. The old list excluded kvm and included lots of extra cruft.
This makes it possible to toggle parallel_emerge just as other scripts
do. In other scripts update the help string to be more specific, the
--jobs option can be used to control parallelism.
Instead of handling toolchain packages in make_chroot and telling
update_chroot to skip the toolchains just depend on update_chroot to do
it properly. Reduces our code duplication by a tiny but worthwhile bit.
When a user creates a chroot and as a common primary group such as
'users' the groupadd command fails. Instead treat this the same as users
and only fail if the group exists but has a different (such as the
'users' group not using GID 100). Hopefully this works better.
If the user already exists check that the UID and GID are correct and
modify it (setting shell and home directory) to match what the SDK
expects. This avoids needlessly failing if the user calling cros_sdk is
the 'core' user on a CoreOS machine.
Change new-user creation to copy the user's full name and group instead
of using a generic name and Google's 'eng' group. Also remove the
default password for the account, it isn't needed and uses perl.
Our SDK tarballs aren't compressed using pbzip2 so there is no advantage
to using pbzip2 to decompress them over bzip2, however lbzip2 does offer
a big advantage. Also trust that the portage config defines a valid
version of bzip2 since we have control over the tarball creation and can
make sure to always include required utilities.
If our sdk has an /etc/mtab file already, then clobber it. This fixes
build problems where chromeos-base now installs /etc/mtab for us, but
the sdk build isn't expecting it leading to the error:
INFO cros_sdk:make_chroot: Running init_setup()...
ln: creating symbolic link `/b/cbuild/new-sdk-chroot/etc/mtab': File exists
Running ['/b/cbuild/src/scripts/sdk_lib/make_chroot.sh', '--stage3_path',
'/b/cbuild/built-sdk.tar.xz', '--chroot', '/b/cbuild/new-sdk-chroot',
'--cache_dir', '/b/cbuild/.cache', '--nousepkg'] failed!
BUG=None
TEST=`cros_sdk --chroot foo` still works
Change-Id: I539cf329e93e28534e6ff00577ce415d76918b85
Reviewed-on: https://gerrit.chromium.org/gerrit/43641
Reviewed-by: David James <davidjames@chromium.org>
Commit-Queue: Mike Frysinger <vapier@chromium.org>
Tested-by: Mike Frysinger <vapier@chromium.org>
As-is all of the various emerge wrapping scripts default to using
--getbinpkg whenever --usepkg is enabled. This means every single emerge
command made makes multiple synchronous HTTP requests to the upstream
binary package repository to get the latest package list. This gets
really frustrating when working remotely with limited network
connectivity. Using --usepkg with --nogetbinpkg will use locally cached
packages without making remote requests.
use the efunctions package for the /etc/init.d/functions.sh script
instead of backing up the old function.sh which doesn't work with the
new baselayout
we remove openrc which provides /etc/init.d/functions.sh. Unfortunatly
other things rely on this file. Stash it away in /tmp/ then restore it
for now.
Change-Id: I18a59e05ecdf08cc8a560b29049c8d25ac1bf5a3
Slipped past during rename of the chroot upgrade script from
49 to 50; name was slightly changed but full re-validation of the
rename wasn't done (thus the typo slipped past testing, and review).
Simplify the code via removal of invoking the upgrade script, instead
just doing the relevant commands (fixing chroot awareness issues in
the process).
BUG=None
TEST=manual cros_sdk invocation
Change-Id: I122de8b4cf7ec0845643e09e7919cbcdbd0bb79a
Reviewed-on: https://gerrit.chromium.org/gerrit/41202
Reviewed-by: Brian Harring <ferringb@chromium.org>
Tested-by: Brian Harring <ferringb@chromium.org>
Rather than having to find /home/${SUDO_USER:-${USER}}/trunk, instead
just look for /mnt/host/trunk (defined by common.sh as $CHROOT_TRUNK_DIR).
This simplifies code flow, and is a requirement for shoving chromite
into PYTHONPATH globally w/in the chroot.
BUG=chromium-os:37347
TEST=cros_sdk --replace; cros_sdk w/ chroot upgrade.
Change-Id: I9ee3e6556541a91193f49cbf74ffc5a8e090537f
Reviewed-on: https://gerrit.chromium.org/gerrit/39921
Tested-by: Brian Harring <ferringb@chromium.org>
Reviewed-by: David James <davidjames@chromium.org>
When running on NFS, the root user may not be able to access ~/.ssh and
~/.gitconfig, so it is necessary to fallback to SUDO_USER to access these
files.
To discourage users from using NFS homedirs, print warnings every time
cros_sdk is run with an NFS homedir.
BUG=chromium-os:36783
TEST=Try cros_sdk --replace and cros_sdk with and without NFS homedirs.
Change-Id: I4cdbceca485d3491656d6f743814da4ebcdd75ad
Reviewed-on: https://gerrit.chromium.org/gerrit/38953
Commit-Ready: David James <davidjames@chromium.org>
Reviewed-by: David James <davidjames@chromium.org>
Tested-by: David James <davidjames@chromium.org>
Explicitly build curl/openssl/git since the toolchain itself tries to
fetch over http with git.
BUG=None
TEST=`cros_sdk --bootstrap` works
TEST=`cbuildbot chromiumos-sdk` works
Change-Id: I50b3145732f8345d6ad6ada41325648cbea31b84
Reviewed-on: https://gerrit.chromium.org/gerrit/36995
Tested-by: Mike Frysinger <vapier@chromium.org>
Reviewed-by: Han Shen <shenhan@chromium.org>
Tested-by: Han Shen <shenhan@chromium.org>
Commit-Ready: Mike Frysinger <vapier@chromium.org>
sudo takes 150ms per invocation on Goobuntu, and with 10 invocations in
enter_chroot.sh, this means that we're wasting a lot of time, every time
cros_sdk is invoked. Cutting these unnecessary invocations reduces the time
required to run enter_chroot.sh from 2.3s to 0.8s.
CL:36618 is the companion change that updates cros_sdk to invoke
sudo unshare -m prior to calling enter_chroot.sh.
Summary of changes:
1. Remove all calls to sudo and just run the commands directly.
- Remove the mount queue and any sudo_multi optimizations.
- Rename sudo_chroot -> bare_chroot because we don't run sudo anymore there.
- Remove code for validating sudo timestamp.
2. Allow the scripts to work as root:
- Ensure that files created by cros_sdk that previously were owned by the
user still are owned by the user (either using chown or cp -p).
- Use $SUDO_USER to find the user's account.
- Use $SUDO_HOME instead of $HOME to find the user's home dir.
- Remove outdated code for disabling automount on Lucid, which doesn't work
when run as root.
- Update code for calculating the user's git username to use sudo to switch
to the user. Also move it to make_chroot.sh so that this change doesn't
impact performance.
3. Cleanup
- Remove environment syncer process in favor of just syncing once when chroot
is entered.
- Remove teardown and instead rely on unshare to unmount the mounts. To make
sure that outside processes never notice the mounts, we use mount -n. This
also ensures that /etc/mtab never contains stale mounts.
- Remove path-overrides, since it is no longer needed.
BUG=chromium-os:35714, chromium-os:35679
TEST=Trybot runs.
CQ-DEPEND=CL:36618
Change-Id: I919a8aadb08fafde97348e8511573c28fdd47186
Reviewed-on: https://gerrit.chromium.org/gerrit/36619
Tested-by: David James <davidjames@chromium.org>
Reviewed-by: Mike Frysinger <vapier@chromium.org>
Commit-Ready: David James <davidjames@chromium.org>
Users sometimes want to run gclient inside the chroot, so we shouldn't
tell users that using it is a bad idea.
The original reason why this message was added is historical: Originally,
users had a newer version of SVN inside the chroot compared to on their
workstation, so if you ran SVN inside the chroot it would permanently upgrade
your working copy such that the version of SVN outside the chroot did not work
with it anymore. This isn't a problem anymore, so we can remove the message.
BUG=none
TEST=Run remote trybot runs of chromiumos-sdk
Change-Id: I7b82a5c94e29d5928f4bb296ae2d99cef397d365
Reviewed-on: https://gerrit.chromium.org/gerrit/36346
Reviewed-by: Mike Frysinger <vapier@chromium.org>
Commit-Ready: David James <davidjames@chromium.org>
Tested-by: David James <davidjames@chromium.org>
Building the chroot environment from sources using
"--bootstrap" currently runs into a circular dependency:
curl->openssl->git->curl
The openssl->git dependency comes indirectly from the fact
that the current version of openssl uses the "cros-workon"
ebuild package to assist in applying packages. The ebuild
system automatically and silently resolves this circular
dependency by reverting the openssl library to an earlier
version that does not use cros-workon based patching.
Unfortunately this older version of openssl has a bug that
causes it not to work when doing builds in a firewalled
environment: When curl (using this older version of openssl
library) attempts to fetch an "https" url, it authenticates
the target server against a bundle of certificate-authority
certificates it maintains. Finding the certificate fails
(although the validation succeeds if curl is told explicitly
what certificate to use). With the certificate not-found,
server authentication fails, the curl download fails, and
the build ultimately fails.
This patch breaks the circular dependency, allowing a
more-current version of openssl to be used in curl, making
the above build scenario work in a firewalled environment.
The circularity is broken by first building git without curl
support (and webdav that depends on curl). Then early
toolchain components up through and including curl are
built. This build of curl then uses a more up-to-date
version of openssl with the desired bug-fix. Once curl is
built, then git is re-built and re-installed with the
now-installed version of curl (re-)enabled.
BUG=None
TEST=create chroot with --bootstrap ; build_packages (behind firewall)
Change-Id: Iaa560fdb6623fcb73cde066a3b2bc2a342169c62
Reviewed-on: https://gerrit.chromium.org/gerrit/34292
Reviewed-by: Mike Frysinger <vapier@chromium.org>
Commit-Ready: paul drews <paul.drews@intel.com>
Tested-by: paul drews <paul.drews@intel.com>
This is forced by cros_sdk; in conjunction w/ this,
drop --distfiles and mangle the chroot on during entrance
dropping a symlink in the old /var/cache/distfiles location
pointing to the new mounted cache_dir location.
Additionally, thread CHROMEOS_CACHEDIR down through the end.
Do this without relying on a version upgrade script- we can't
require they be run before entering, thus we exploit the fact
that cros_sdk explicitly forces a write lock to do the upgrade,
if we see the old form we know we can do the upgrade w/out
worrying about collisions.
CQ-DEPEND=CL:33871
BUG=chromium-os:34457
TEST=manual testing.
Change-Id: I6805266e3ec683f05d3ba615f9e8840642a28e48
Reviewed-on: https://gerrit.chromium.org/gerrit/33868
Commit-Ready: Brian Harring <ferringb@chromium.org>
Reviewed-by: Brian Harring <ferringb@chromium.org>
Tested-by: Brian Harring <ferringb@chromium.org>
